Research On Coordinate Measurement Technology Based On Binocular Stereo Vision

Binocular stereo vision is an important branch of computer vision,which has high application value in military and civil fields.Based on the existing camera calibration algorithms,a binocular camera calibration algorithm is proposed in this paper.A binocular stereo vision tracking and measuring system is constructed by combining binocular stereo vision measurement with image tracking technology.Problems of binocular camera calibration methods still exist,such as the complicated nonlinear optimization steps,the couples of the camera internal,external parameters and lens distortion,and the relatively large time cost.In view of this,in this paper,a calibration method for binocular camera is proposed,which combines distortion correction with homography.In the undistorted image,the vector of any two points in the straight line should be zero when the direction of the vector is simultaneously.Based on the characteristics,the distortion coefficient of binocular camera can be solved,and the measurement matrix can be constructed through homography,then the initial value of the camera inside and outside parameters can be obtained by matrix decomposition,and finally the parameters of binocular camera are obtained by nonlinear optimization.The simulation and experimental results show that the accuracy of this method is similar to that of the traditional calibration method,but the calculation time is shorter,and the calibration efficiency is better than the traditional calibration method.It provides a new idea for the calibration of binocular camera.In the image tracking algorithm,the tracking of moving objects is generally two-dimensional and lacks depth information.But in some cases,3d information of moving objects is required.In this paper,a set of binocular stereo vision measurement and tracking system is constructed based on VS2013,with binocular stereo vision measurement and image tracking algorithm are combined.In the left and right camera image,the moving object is detected by Vibe foreground detection algorithm,and the moving object tracking is achieved by KCF tracking algorithm,and then the parallax figure is got by object stereo matching of the left and right camera,then finally the 3d information of moving objects is obtained.By comparing the error of theoretical depth and the actual depth between the moving object and the binocular camera,the feasibility of the system is verified...